The proportions shown in the last column above represent the 

 net effective proportions reflecting the different sampling rates, 

 the reweighting factors applied to offset these different sampling 

 rates, and the different connpletion rates achieved. The latter 

 are shown in the following table: 



Completion 

 City Rate 



% 



Atlanta, Georgia 96 



Chicago, Illinois 89 



Cleveland, Ohio 80 



Denver, Colorado 87 



Houston, Texas 81 



Los Angeles, California 92 



New York, New York 82 



Omaha, Nebraska 91 



Portland, Oregon 83 



Springfield, Massachusetts 86 



While the sample was designed so that the "expectation" (based 

 on the available data) would be of a subsample , adequate for 

 separate analysis, for each of four subuniverses , these "expec- 

 tations" were not always realized. In some cases, the samples 

 produced for certain subuniverses were too small for separate 

 analysis. In such cases, subuniverses No. 1 and No. 4 were 

 merged, and/or subuniverses No. 2 and No. 3, the first two 

 representing, roughly, the "public" eating places ajid the latter 

 two the "institutional". On this merged basis, adequate samples 

 for analysis were obtained in all cities. 



While the list sample was a single-stage sample, the area sam- 

 ples in all cities except Springfield, Massachusetts, were 

 two-stage sannples. The first stage was a sampling of areas, 

 as explained above. The second stage was a sannpling of the 

 establishments listed by the field staff in their full canvas of the 

 sampled areas . In Springfield, Massachusetts, the area sample, 

 too, was a single-stage saimple, because, in the first stage, the 



